Pleasantly surprised to see a project like this without genAI nowadays.
And for people inevitably asking "why not Blender," my answer is that the field of 3D content creation is so vast that some people need "gateway drug" like this (and SketchUp, and Daz3D, etc) before trying to throw themselves into Blender/Maya/Houdini's throats.
As a dev trying to create 3D games on my spare time, I actually am (I'm also eager to use stableprojector once I upgrade my 10yo pc)
It's really hard having to be a 3D modeler, animator, art director, developer and having good enough taste to know you're never going to have the time to build what you find acceptable
First time making something 3D and code wise, codex has been very useful, even creating a in-game 3D editor as well as netcode
The trickiest part is really using 3D and it comes with lot of extra scoping you normally take for granted: animation, uv texture, rigging for humanoids, making sure stuff doesn't clip through etc.
Still learning Blender but its very slow. I haven't tried the MCP for it yet but I want to get proficient at it to be able to produce psx graphic models and textures...
This person related that in the years prior to this one, artists were essential members of every Game Jam team. Getting art for the games was the hardest, most limiting step. Without an artist, the games were limited in scope. Not every team could get an artist.
This year, many teams didn't have artists at all, and yet they were able to create more complete games with good looking art assets.
This isn't "firing" artists. This is making engineers more effective.
I work in the video space now, and I can't even begin to tell you how effective AI models make editors. It's like a super power.
Very nice project!
Thanks for making it open-source and very easy to use. I hope you'll find contributors and be able to get a nice collection of animations!
I'm starting a new project, I may have to use this tool in the next few months.
I can see this being huge for 3d-printing. Apply on a sculpted .stl, reposition and download/save the new pose. There is already stuff in Blender to do stuff like that, but Blender is almost like a hobby in itself, just like the actual printing is like a hobby (to learn and master).
This and more mocap's plz. Keep going! The multiple animation export to glb saves me a ton of time. This is great! Combine this with ArmorPaint and you're almost there. Blender is great but it's so heavy.
This is cool! I was confused how to get going at first and didn't realise it was a step-by-step wizard I had to follow, plus I only watched the demo video briefly. I loaded the human then clicked around the main area and couldn't get anything to happen, and almost gave up.
Suggestion to help with this: Instead of starting with a blank slate, start right at the end of the process with the human model loaded and animated, then the user can go "back" and change things once they've seen the kind of results they can get.
The next release I have for this has an "explore" page to start. What exists now is more of a 'build your own' page...which isn't great when visiting for the first time. See this small video of the new UX.
Having the skeleton be auto-deduced (through the ML or otherwise) would've been a very natural fit here as a feature. Knowing that it's basically possible to do, selecting the skeleton type and then adjusting it manually, joint by joint, seems ... archaic?
Why the condescending tone, OP has a point. Techniques like RigAnything[0] exist - whether this is within the scope of this project is another topic, but this seems like a reasonable request to me (especially considering how many x_2_y-named projects are ML related)
can you give an example for clarification of what you are wanting to do? This web app is pretty flexible, so other types of rigs or animations could be added?
That’s actually a really interesting take (even though the OP software is not really for that). How would you _like_ to design this treehouse? Is the particular tree relevant here (do you have a photogrammetric scan of it for insrance) or are you planning to design a generic shape you are confident can be installed on the target trunk? Is the treehouse more of a plank platform for one or an actual multiroom house? Is the budget closer to 5$ or 50 000$?
This is great! I've been using mixamo and really sion, but would just love something open source that does both the body rigging as well as the facial rigging and lip syncing. It'd make life so much easier
All in typescript too. Actually very impressive. Well, some webm videos and .glb 3dfiles but only the essentials it seems, the rest is all propper typescript.
Pleasantly surprised to see a project like this without genAI nowadays.
And for people inevitably asking "why not Blender," my answer is that the field of 3D content creation is so vast that some people need "gateway drug" like this (and SketchUp, and Daz3D, etc) before trying to throw themselves into Blender/Maya/Houdini's throats.
Given the name I was sure this was going to be AI.
genAI doesn't seem to scale very well without large costs and subscriptions to pay for it. Doesn't seem good for an open source project.
why? aren't you desperately longing for trying to put into a prompt how a model should be animated? /s
As a dev trying to create 3D games on my spare time, I actually am (I'm also eager to use stableprojector once I upgrade my 10yo pc)
It's really hard having to be a 3D modeler, animator, art director, developer and having good enough taste to know you're never going to have the time to build what you find acceptable
First time making something 3D and code wise, codex has been very useful, even creating a in-game 3D editor as well as netcode
The trickiest part is really using 3D and it comes with lot of extra scoping you normally take for granted: animation, uv texture, rigging for humanoids, making sure stuff doesn't clip through etc.
Still learning Blender but its very slow. I haven't tried the MCP for it yet but I want to get proficient at it to be able to produce psx graphic models and textures...
I saw a post by a Gam Jam organizer recently.
This person related that in the years prior to this one, artists were essential members of every Game Jam team. Getting art for the games was the hardest, most limiting step. Without an artist, the games were limited in scope. Not every team could get an artist.
This year, many teams didn't have artists at all, and yet they were able to create more complete games with good looking art assets.
This isn't "firing" artists. This is making engineers more effective.
I work in the video space now, and I can't even begin to tell you how effective AI models make editors. It's like a super power.
Very nice project! Thanks for making it open-source and very easy to use. I hope you'll find contributors and be able to get a nice collection of animations!
I'm starting a new project, I may have to use this tool in the next few months.
Could also check out the Auto Rig Pro ( https://artell.gumroad.com/l/auto-rig-pro ) plugin in Blender, as it does corrective shape-keys and makes exports to Unity/Unreal trivial. Also, FaceIt ( https://faceit-doc.readthedocs.io/en/latest/ ) uses the iPhone Lidar with the Unreal App, and makes facial expression capture less buggy.
Keep in mind most motion capture files are considered live-performances, and must be licensed by the original rights holders:
https://www.mocapclub.com/Pages/Library.htm
https://github.com/MattiasFredriksson/io_anim_c3d/releases
This project sort of works for basic mocap (calibration needs ChArUco board):
https://freemocap.org/
Or if you have the resources and athletic actors around:
https://www.rokoko.com/
Or for student projects could try:
https://www.mixamo.com/
Hope you make something awesome and fun =3
I can see this being huge for 3d-printing. Apply on a sculpted .stl, reposition and download/save the new pose. There is already stuff in Blender to do stuff like that, but Blender is almost like a hobby in itself, just like the actual printing is like a hobby (to learn and master).
What algorithm is it using for applying weights? Blender uses automatic bone heat weighting, but sometimes it can fail.
There's an open-source plugin that does surface-based heat diffusion and also this really clever voxel solution: https://superhivemarket.com/products/voxel-heat-diffuse-skin...
Yessssssss......
This and more mocap's plz. Keep going! The multiple animation export to glb saves me a ton of time. This is great! Combine this with ArmorPaint and you're almost there. Blender is great but it's so heavy.
Great job.
This is cool! I was confused how to get going at first and didn't realise it was a step-by-step wizard I had to follow, plus I only watched the demo video briefly. I loaded the human then clicked around the main area and couldn't get anything to happen, and almost gave up.
Suggestion to help with this: Instead of starting with a blank slate, start right at the end of the process with the human model loaded and animated, then the user can go "back" and change things once they've seen the kind of results they can get.
The next release I have for this has an "explore" page to start. What exists now is more of a 'build your own' page...which isn't great when visiting for the first time. See this small video of the new UX.
https://x.com/scottpetrovic/status/1980697224142106716
> Supports humanoid, four-legged, and bird creatures.
Man is a featherless biped
- tantalor
This is defo a well needed project. The only other alternative I can think about is Mixamo. Goodluck and hope this projects overtakes Mixamo
I just wanted to thank the creators of this app for making it usable without having to create an account. It's a breath of fresh air!
Yes. I really have sign up and subscription fatigue, so I don't want any of that happening. There is too much of that on the web already.
Having the skeleton be auto-deduced (through the ML or otherwise) would've been a very natural fit here as a feature. Knowing that it's basically possible to do, selecting the skeleton type and then adjusting it manually, joint by joint, seems ... archaic?
Have you done 3D modelling/rigging/animating yourself before?
Why the condescending tone, OP has a point. Techniques like RigAnything[0] exist - whether this is within the scope of this project is another topic, but this seems like a reasonable request to me (especially considering how many x_2_y-named projects are ML related)
[0] https://www.liuisabella.com/RigAnything/
amazing timing! been needing something like this since Blender has been tough.
I've been working on a cybercafe simulator with Three.js and codex and animation is hard to get right
https://x.com/AgentifySH/status/1981761490274464068
Great project! It uses the motion library from the artist quaternius. He has some wonderful free and low-cost models.
nice, animating spinning wheels or tracks on vehicles would be cool!
can you give an example for clarification of what you are wanting to do? This web app is pretty flexible, so other types of rigs or animations could be added?
OP might be thinking about something like this paper "Rig My Ride: Automatic Rigging of Physics-based Vehicles for Games" by Katz et al.
[0] https://melissa2661.github.io/RigMyRide/
i need to map out tree house plans - can this be used?
I'm sorry I can't quite parse what you are asking.
"map out" - do you mean you want to digitally design tree houses?
yes, kind stranger. i need to design tree house, was not sure if this can be used. not so fond of sketch up.
This software is for animating preexisting 3d models, not creating new ones from scratch, so no, it will not serve to model a tree house.
That’s actually a really interesting take (even though the OP software is not really for that). How would you _like_ to design this treehouse? Is the particular tree relevant here (do you have a photogrammetric scan of it for insrance) or are you planning to design a generic shape you are confident can be installed on the target trunk? Is the treehouse more of a plank platform for one or an actual multiroom house? Is the budget closer to 5$ or 50 000$?
No, use sketchup
This is great! I've been using mixamo and really sion, but would just love something open source that does both the body rigging as well as the facial rigging and lip syncing. It'd make life so much easier
All in typescript too. Actually very impressive. Well, some webm videos and .glb 3dfiles but only the essentials it seems, the rest is all propper typescript.
But a very nicely put together repo. Good job.